City Research Online

Heuristics and multi-dimensional physical database design

Fu, Z. (1991). Heuristics and multi-dimensional physical database design. (Unpublished Doctoral thesis, City, University of London)


An expert system approach has recently been used in parameter selection for VSAM (Virtual Storage Access Method) file organisation [AL87a]. This system has been developed to aid in-house users to apply relevant facts and heuristics to optimise VSAM file design. Multi-dimensional physical
database design is more sophisticated and complicated than VSAM file design. The expert system approach can be applied to select and tune physical database design for various applications.

A great deal of work has been done in developing diverse algorithms or access methods to organise automated information on secondary storage devices [FA86b] [FR86] [FR88] [GU84] [HU88a] [KS88a] [KS86] [L087] [NI84] [OR88b] [OR86] [OT85] [R081], etc. However, little work has been done to enable designers to select an access method which matches a projected application profile (features and requirements) and perceived strengths and weaknesses of candidate algorithms. This thesis considers a number of grid based algorithms and makes expert assessments of each according to its strengths and weaknesses. It analyses features of various access methods and using expert knowledge matches features for a range of m-d (multi­ dimensional) algorithms with corresponding characteristics of an application. The knowledge-based system presented in this thesis can be applied either manually or computerised to give a systematic approach to m-d algorithm selection. A system is proposed to (1) heuristically select an initial algorithm; (2) describe how the selection process is evaluated against actual m-d algorithm performance and (3) show how the results of the evaluation can be used to refine expert knowledge embodied in the selection system. Heuristic assessments are given for several m-d access algorithms. Examples are
presented to show how these heuristics are used to select a m-d access algorithm for a specific application. It is reasonable to suppose that the initial heuristic assessments are not entirely accurate. A tuning mechanism for the system heuristics is given in section 4.9. The system selection process is thereby, able to adjust to real world results. Finally, we present a simple example to illustrate how the proposed system works.

Publication Type: Thesis (Doctoral)
Subjects: Q Science > QA Mathematics > QA75 Electronic computers. Computer science
Departments: Bayes Business School > Bayes Business School Doctoral Theses
Bayes Business School
Doctoral Theses
Text - Accepted Version
Download (9MB) | Preview



Downloads per month over past year

View more statistics

Actions (login required)

Admin Login Admin Login